On Mon, 23 Aug 2010, João Martins wrote:
The usual reply to myself kind of thing. I've just tested an earlier pd-extended package, with GEM 0.91.3 (that's the one working in my Mac), and the problem persists. When I create a "pool" instance and open the help patch and then try to close it (this is the fastest way to test, since it produces the same output as putting a fid_class object into the patch), this is what I get at the terminal console:
If you can be sure that the problem comes from pool, but the error message doesn't, then you can't rely on the glibc backtrace, nor on a gdb backtrace either. You have to use valgrind, which can do early diagnostic, instead of gdb and glibc that can only tell you the last step of the death.
But I don't recall whether it works on OSX, or only Linux.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard on the road (in Îles-de-la-Madeleine, QC).